Cognitive Load Measurement And Its Application In Multimedia Learning

After Cognitive Load Theory (CLT) was proposed by John Sweller in1980s',researchers began to invent new methods for measuring cognitive load. Along with thedevelopment of cognitive load theory, a fundamental challenge for itself is how tomeasure cognitive load. At present, the measurement methods can be devided into threecategories, namely, subjective measures, task performance measures and physiologicalmeasures. Because of its simplicity, practicability and convenience, subjective measurewas the most popular in recent influential researches on cognitive load. In addition, lotsof research effort have been donated on the effects of task characteristic andinstructional design on cognitive load, while the relationship between learner'scharacteristic or learning mode and cognitive load have been neglected. Using manypopular research methods in psychology, including investigation technique, measuringmethod, experimental method, the present study was designed by combining a series ofexperiments to investigate the following phenomenon in CLT. Firstly, by comparing thesensitivity and validity of three subjective rating scales, the most effective one of thethree was identified. Secondly, using this identified scale, the effects of the learnercharacteristic and learning mode on the cognition load and learning performance in themultimedia learning environment were probed. Finally, based on the above mentionedresults and previous literature, a model on how learner's characteristics and learningmode affect cognition load and learning performance was proposed. The study providedimportant basis for the cognitive load measurement in the multimedia learning,instructional design and personalized learning.The main results of the research were:1. In the present task condition, the sensitivity, validity and diagnosticity of theWorkload Profile Index Ratings(the WP)were higher than the Paas Cognitive LoadScale(the PAAS), while the ones of the NASA Task Load Index(the TLX)were theworst. Base on the above mentioned results, the WP was the most effective subjective rating scales of the three ones, especially for medium or lower difficulty task.The resultsalso showed that the sensitivity and intrusiveness of response time of secondary-taskwere both better and it could be used as reference object to subjective measures. Thesensitivity of cognitive load's composite index was also better.2. In the present research condition, the generalization from educator reduced thelearner's intrinsic cognitive load, and increased their recall and transfer test performance;while the generalization from learner increased their germane cognitive load, butincreased their transfer test performance only. The learning performance of thegeneralization from educator was better than that of the generalization from learner andnot generalization, either for difficult materials or for easy ones.There was interactioneffect on cognition load and learning performance between generalization style andmaterial difficulty.3. In the present research condition, meta-cognitive questioning had significantinfluence on cognitive load and learning performance,for it increased meta-cognitiveload and total amount of cognitive load, and it can also increased the learner's learningperformance. For the complex learning materials, cognitive load from meta-cognitivehinting questions was higher significantly than one from meta-cognitive presentingquestions. There was interaction effect on cognition load and learning performancebetween meta-cognitive questioning and material difficulty.4. In the present research condition, learning style and material presentation modehad significant influence on cognitive load, while there was interaction effect betweenlearning style and material difficulty, but there was no interaction effect betweenlearning style and material presentation mode. For the complex learning materials,cognitive load from the learner of visual learning style was higher significantly than onefrom the learner of verbal learning style, and cognitive load from full-text presentationwas higher significantly than one from summarizing presentation. The learning styleand material presentation mode had no significant main effect on the learner's learningperformance, while there was interaction effect on transfer and recall test performancebetween them, except for recognition one.5. Based on the above mentioned results and previous literature, a model on howlearner's characteristics and learning mode affect cognition load and learningperformance was proposed. The model showed that, learning mode (i.e. generalization or meta-cognitive questioning) affected effective cognition load, including germanecognitive load and meta-cognitive load, while learning style affected affected intrinsiccognitive load; interior factors affected germane cognitive load,meta-cognitive load andintrinsic cognitive load, while external factors affected intrinsic and extraneouscognitive load. In addition to, there was interaction effect on cognition load and learingperformance between interior factors and external ones.